RSS and Atom are open formats exchanged via open protocols within an open eco-system of interworking distributed servers. Twitter is a company providing a centralized service accessed via proprietary protocols. Open distributed systems inevitably scale better than proprietary centralized systems. What microblogging needs is not a centralized service but an open protocol.
bob wyman
dave
· 3 months ago
Thanks Bob. Couldn't agree more.
mariva
· 3 months ago
Well stated.
technogran
· 3 months ago
love this post and couldn't agree more! It amazes me how others haven't tumbled to just how useful RSS feeds are, and as you say, they never let you down. I use RSS all the time, to bring me my news, any blog posts I am interested in, and I have used it as you do, in combination with Twitter
martin_english
· 3 months ago
It's probably due to my style of work and / or workflow, but I can't function without RSS - An appropriate example is that many of the people I follow on twitter are outside my timezone. So I follow their notifications / conversations via RSS.
Having some or most of my incoming signals as RSS means any other signal must also be RSS compatible. If I can't read you directly via RSS, you better be worth the effort for me to go to one of those services that produce an RSS feed, or I dont bother.
(PS I'm not into format wars, whatever the location - When I say RSS feed I mean anything that my RSS reader can parse - be it RSS, ATOM, or any other XML like 'thingo')
adario strange
· 3 months ago
nice! the title says it all.
ann
· 3 months ago
That's what I'd like to see: Dave on Oprah...
Mark Essel
· 3 months ago
Enjoyed your comment on Freds post, and even more so your explanation here Dave. I asked you there about a minimum update time for rss. Is it really no faster than one minute (due to the nature of the protocol)?
Are there alternative architectures that are distributed buy faster to respond?
As a pretty active user of web info, my ideal would be distributed but instantly updated. Too crazy to hope for?
dave
· 3 months ago
Mark there are always tradeoffs in tech.
That said, there may be ways to do it, the banking industry and stock markets obviously cracked this nut a long time ago.
But I always like to underpromise and overdeliver. That's kind of the way an engineer's mind works. Leads to happier users. :-)
Mark Essel
· 3 months ago
Without Scotty's wise delivery method from star trek we would all be in grave peril of the overexpecting user syndrome.
I rather like pubsubhubbub's approach to this. By my understanding, pubsubhubbub has the hub parse the feed and send the new or modified items in it along with the ping, so that (in the common case, at least) the client doesn't need to fetch the feed at all. It adds some complexity to the hub, but the savings in bandwidth would be significant.
dave
· 3 months ago
Makes no difference, imho -- parsing feeds is well-known art.
Believe it or not at UserLand we did the exact same thing at the very beginning of RSS, with the exact same rationale. Guess what, the other developers ignored it and wrote their own feed parsers.
Now nine or ten years later, there's even less reason to worry about this. Just creating another format and another protocol -- and that's the last thing the world needs.
That said, it's no big deal. If people want PSHB, go for it. It's an upgrade for RSS too.
sull
· 3 months ago
right. it's important to acknowledge pshb and inspect how they are approaching this but i also tend to prefer to stick with what already exists whenever possible. that's just me. so rsscloud is prior art to pshb (by many years) and so its rsscloud that i'm focusing on. but they will co-exist and compliment the effort of loosely coupled de-centralized near real-time pub/sub/agg.
with that, there's no saying that rsscloud wont provide similar data distribution methods. i've experimented with this myself and it also dates back to the origins of the commentAPI where you send rss item data chunks (xml) to the remote script over http.
Mark Essel
· 3 months ago
I'd heard about some promised fast response from phsb, and even talked through a similar concept on an old friendfeed dev post before I realized it already had been implemented.
Looking forward to any formats that work with distributed communication systems
adrianeden
· 3 months ago
Utilizing RSS Feeds can be extremely powerful, especially with Yahoo Pipes and pubsubhubbub. They can also be used to strengthen your SEO. I love RSS Feeds.
Anthony Farrior
· 3 months ago
RSS is the current vehicle content rides on. It's the reason why we have news instantly before tv does. Twitter can't sub for RSS b/c 90% of the time content put on twitter and digg manually came from someone staring at their automatic rss feeds trying to be "first"...imho
dave
· 3 months ago
You got it babe, that's exactly right.
Steve Farrell
· 3 months ago
RSS is scalable in a technical sense, but is it scalable in a human sense? Web pages have hyperlinks, from which search engines can infer value... RSS feeds lack a analogous property that would allow systems to rank entries. As a result, they're not terribly useful by themselves.
sull
· 3 months ago
in the case of human interaction, RSS is meant to be parsed into readable formats like html. else, you can transform RSS using XSL, CSS and HTML. to demonstrate the latter, view this RSS feed in Firefox - http://go.vocal.ly/1k - view source shows you pure RSS.
also, search engines have complex algorithms and utilize RSS feeds to aid in content discovery.
as far as inferred link value, can you ellaborate?
Steve Farrell
· 3 months ago
By inferred value, I mean something like pagerank. There are reproducible, systematic ways to rank web pages based on the link structure of the web. I don't see anything like that for RSS.
As a result, I have no way to know which of the 500 news items in my reader are worth my time: that's what I mean by human scalability.
dave
· 3 months ago
1. It has a link element, both at the channel level and the item level.
2. The description in an item can have HTML markup so it has exacctly the same capability as HTML.
#2 pretty much makes this moot. Please, either continue this on your own blog or let it go. Thanks.
When Twitter can't even serve up it's own tweets from more than a week ago there is no hope of it consuming RSS.
I pull any search term I am interested in on Twitter as an RSS feed and drop it in to my Google Reader account. That way I can maintain my own history. What I should probably do is re-share those streams publicly. If we all did that it would pull Twitter in to the RSS cloud - whether Twitter likes that idea or not.
Henri Asseily
· 3 months ago
And then you merge RSS with DNS and you've got a solid base from which to publish your rss feed urls. "dig rss.henri.tel naptr" That will show you my rss feeds, from a static, single unchanging access point. Best of all worlds.
dave
· 3 months ago
One step at a time. :-)
Matej Cepl
· 3 months ago
Of course you meant
host -t SRV _feed._tcp.example.com
right? ;)
William Mougayar
· 3 months ago
Dave, Just for the record, I was the one that suggested that button on Fred's comments, and Richard seconded that thought, then he added his other comment later. (if you scroll-up, you'll see that).
I was suggesting it merely as a convenience for certain cases, but not as a way to totally subsume or inundate Twitter. Since 65% of Twitter streams is apparently coming from bots- this is already happening either via one of the RSS-to-Twitter feeder (which are not very reliable) or via a direct API feed.
dave
· 3 months ago
You're absolutely right -- and I'm sorry for the confusion. I've corrected the post.
I even see the comment, but I can't find a permalink to it, so I pointed to one of yours in that thread that has an easy to find permalink.
Anyway, if Twitter were to become an RSS aggregator where anyone could give it any feed, I think they'd quickly hit an even bigger scaling wall than any they've hit so far. One of the reasons Twitter works is because even though they have a huge number of users, there's a cap on the variety of things you can follow. With RSS there is no cap.
For example, today I told my aggregator to check every 10 minutes on Mininova for a BBC video from Misano about Motogp. I'm looking for something very specific, and it will show up there sometime in the next 24 hours. As soon as it is, I want to download it.
Problem is there are millions of possible queries you can do on Mininova, and that's just one site. If any significant portion of their users were doing this, they'd be checking millions of feeds and having the same problems Google is having with Google Reader and Feedburner, without the revenue stream Google has.
It's a nice idea -- I'm doing stuff like that myself with twitter: http://twitter.com/friendsofdave -- but here's the key point -- my computer is doing the feed scanning, not theirs. That's why they can afford to do it.
With the disclaimer that I don't have access to their data, so it's basically conjecture.
William Mougayar
· 3 months ago
(Not a big deal, but Thanks for the correction) I agree that Twitter would flaunch and bloat if it were to consume more RSS feeds. Actually, I would be in favor of limiting the # of allowed Tweets per day, and charging businesses that want to send RSS rivers of news into it. That would automatically filter out the non essential bable. I'm not sure what the ave# of tweets/day/user is, but even if it was set to 50, it wouldn't affect 99% of users. Just my 2 cents on that.
I currently follow /friendsofdave from @wmougayar, and also like the Favorites features on Twitter, but it's a pull, not push unfortunately. Have you seen this: http://ctwittlike.appspot.com/ where you enter someone's Twitter ID and you basically snoop of their feeds.
I'm going to play with River2 and rssCloud this week-end, and might have more meaningful feedback, but I was an old user of radio userland & your opml editor almost since day 1, and am a big RSS junkie.
hardaway
· 3 months ago
Caveat: I know nothing here beyond what the ordinary user knows. I can't believe RSS would go away when so many people are JUST discovering they can read blogs and newspapers and magazines through RSS feeds. I think the average Josephine would see no comparison between RSS and Twitter and see them as having totally different purposes. I have long read some tweeters through RSS because I want to make sure I see them in my barrage of tweets. So if I were to weigh in, I'd be saying RSS would subsume Twitter. But I'm only amusing myself when I venture into these highly technical discussions.
dave
· 3 months ago
Amuse yourself, that's all the doomsayers are doing.
optionshiftk
· 3 months ago
I get really irked when I see these "tech journalists" parading around and proclaiming "RSS is Dead" when they seem to not grasp the very concept of RSS. As you've said Dave, a good portion of them confuse RSS with Google Reader.
dave
· 3 months ago
It's quite possible that they're not confused. Saying "Google Reader is Dead" could be a bad career move for them. No more one-one-interviews with Eric Schmidt. No more sponsorship for the next conference. Much easier to take your cheap shots at a free and open format that never gives interviews and will never buy an ad or sponsor a conference.
Antonio Ognio
· 3 months ago
When I first heard of microblogging my first thought was "oh it's just creating very short blog posts and aggregating all of the RSS feeds into one stream". I would have been just great if Twitter had just provided the feed discovery and aggregating services since day one just letting all the other blogging platforms create the content.
dave
· 3 months ago
It's not too late for them to do just that! :-)
Joshed
· 3 months ago
Indeed.
Matej Cepl
· 3 months ago
Which is the reason why I really think identi.ca (or laconi.ca or status.net or whatever is its name at the moment) is the right way to go. Only if they had better XMPP interface (which is of decentralized as well).
All Comments
bob wyman
Having some or most of my incoming signals as RSS means any other signal must also be RSS compatible. If I can't read you directly via RSS, you better be worth the effort for me to go to one of those services that produce an RSS feed, or I dont bother.
(PS I'm not into format wars, whatever the location - When I say RSS feed I mean anything that my RSS reader can parse - be it RSS, ATOM, or any other XML like 'thingo')
Are there alternative architectures that are distributed buy faster to respond?
As a pretty active user of web info, my ideal would be distributed but instantly updated. Too crazy to hope for?
That said, there may be ways to do it, the banking industry and stock markets obviously cracked this nut a long time ago.
But I always like to underpromise and overdeliver. That's kind of the way an engineer's mind works. Leads to happier users. :-)
Believe it or not at UserLand we did the exact same thing at the very beginning of RSS, with the exact same rationale. Guess what, the other developers ignored it and wrote their own feed parsers.
Now nine or ten years later, there's even less reason to worry about this. Just creating another format and another protocol -- and that's the last thing the world needs.
That said, it's no big deal. If people want PSHB, go for it. It's an upgrade for RSS too.
with that, there's no saying that rsscloud wont provide similar data distribution methods.
i've experimented with this myself and it also dates back to the origins of the commentAPI where you send rss item data chunks (xml) to the remote script over http.
Looking forward to any formats that work with distributed communication systems
else, you can transform RSS using XSL, CSS and HTML.
to demonstrate the latter, view this RSS feed in Firefox - http://go.vocal.ly/1k - view source shows you pure RSS.
also, search engines have complex algorithms and utilize RSS feeds to aid in content discovery.
as far as inferred link value, can you ellaborate?
As a result, I have no way to know which of the 500 news items in my reader are worth my time: that's what I mean by human scalability.
2. The description in an item can have HTML markup so it has exacctly the same capability as HTML.
#2 pretty much makes this moot. Please, either continue this on your own blog or let it go. Thanks.
I pull any search term I am interested in on Twitter as an RSS feed and drop it in to my Google Reader account. That way I can maintain my own history. What I should probably do is re-share those streams publicly. If we all did that it would pull Twitter in to the RSS cloud - whether Twitter likes that idea or not.
"dig rss.henri.tel naptr"
That will show you my rss feeds, from a static, single unchanging access point.
Best of all worlds.
host -t SRV _feed._tcp.example.com
right? ;)
Just for the record, I was the one that suggested that button on Fred's comments, and Richard seconded that thought, then he added his other comment later. (if you scroll-up, you'll see that).
I was suggesting it merely as a convenience for certain cases, but not as a way to totally subsume or inundate Twitter. Since 65% of Twitter streams is apparently coming from bots- this is already happening either via one of the RSS-to-Twitter feeder (which are not very reliable) or via a direct API feed.
I even see the comment, but I can't find a permalink to it, so I pointed to one of yours in that thread that has an easy to find permalink.
Anyway, if Twitter were to become an RSS aggregator where anyone could give it any feed, I think they'd quickly hit an even bigger scaling wall than any they've hit so far. One of the reasons Twitter works is because even though they have a huge number of users, there's a cap on the variety of things you can follow. With RSS there is no cap.
For example, today I told my aggregator to check every 10 minutes on Mininova for a BBC video from Misano about Motogp. I'm looking for something very specific, and it will show up there sometime in the next 24 hours. As soon as it is, I want to download it.
Problem is there are millions of possible queries you can do on Mininova, and that's just one site. If any significant portion of their users were doing this, they'd be checking millions of feeds and having the same problems Google is having with Google Reader and Feedburner, without the revenue stream Google has.
It's a nice idea -- I'm doing stuff like that myself with twitter: http://twitter.com/friendsofdave -- but here's the key point -- my computer is doing the feed scanning, not theirs. That's why they can afford to do it.
With the disclaimer that I don't have access to their data, so it's basically conjecture.
I agree that Twitter would flaunch and bloat if it were to consume more RSS feeds.
Actually, I would be in favor of limiting the # of allowed Tweets per day, and charging businesses that want to send RSS rivers of news into it.
That would automatically filter out the non essential bable. I'm not sure what the ave# of tweets/day/user is, but even if it was set to 50, it wouldn't affect 99% of users. Just my 2 cents on that.
I currently follow /friendsofdave from @wmougayar, and also like the Favorites features on Twitter, but it's a pull, not push unfortunately.
Have you seen this: http://ctwittlike.appspot.com/ where you enter someone's Twitter ID and you basically snoop of their feeds.
I'm going to play with River2 and rssCloud this week-end, and might have more meaningful feedback, but I was an old user of radio userland & your opml editor almost since day 1, and am a big RSS junkie.